Test AI-agent interactions with tools, memory, and external systems under realistic failures and security boundaries.
Design bounded AI-agent tools with typed arguments, least privilege, idempotency, and approval gates.
Assesses repository AI-agent configuration and safety guardrails when onboarding or auditing OpenCode, Copilot, or Claude Code setup.
Build MCP servers with typed tools, resource schemas, authorization boundaries, and reliable transport behavior.
Develop RAG retrieval pipelines with provenance, access filtering, chunking, and grounded answer evaluation.
Reviews AWS Organizations, Control Tower, and landing-zone governance when a multi-account AWS assessment is requested.
Reviews Azure workload and landing-zone architecture against Well-Architected and CAF guidance when design changes are requested.
Generates validated first-draft Azure Bicep, ARM, Terraform, or Pulumi when new Azure infrastructure code is requested.